Kenji Fukui

Proteome-wide identification of lysine succinylation in thermophilic and mesophilic bacteria

Lysine succinylation, one of post-translational acylations conserved from eukaryotes to bacteria, plays regulatory roles in various cellular processes. However, much remains unknown about the general and specific characteristics of lysine succinylation among bacteria, and about its functions different from those of other acylations. In this study, we characterized lysine succinylation, a newly discovered widespread type of lysine acylation in five bacterial species with different characteristics such as optimal growth temperature and cell wall structure. This study is the first to demonstrate that succinylation is general phenomenon occurring not only in mesophiles but also in thermophiles. Mapping of succinylation sites on protein structures revealed that succinylation occurs at many lysine residues important for protein function. Comparison of the succinylation sites in the five bacterial species provides insights regarding common protein regulation mechanisms utilizing lysine succinylation. Many succinylation sites were conserved among five bacteria, especially between Geobacillus kaustophilus and Bacillus subtilis, some of which are functionally important sites. Furthermore, systematic comparison of the succinyl-proteome results and our previous propionyl-proteome results showed that the abundance of these two types of acylations is considerably different among the five bacteria investigated. Many succinylation and propionylation events were detected in G. kaustophilus, whereas Escherichia coli and B. subtilis exhibited high succinylation and low propionylation; low succinylation and high propionylation were identified in Thermus thermophilus, and low succinylation and propionylation were observed in Rhodothermus marinus. Comparison of the characteristics of lysine succinylation and lysine propionylation suggested these two types of acylation play different roles in cellular processes.

Structural Features and Functional Dependency on β-Clamp Define Distinct Subfamilies of Bacterial Mismatch Repair Endonuclease MutL

In early reactions of DNA mismatch repair, MutS recognizes mismatched bases and activates MutL endonuclease to incise the error-containing strand of the duplex. DNA sliding clamp is responsible for directing the MutL-dependent nicking to the newly synthesized/error-containing strand. In Bacillus subtilis MutL, the β-clamp-interacting motif (β motif) of the C-terminal domain (CTD) is essential for both in vitro direct interaction with β-clamp and in vivo repair activity. A large cluster of negatively charged residues on the B. subtilis MutL CTD prevents nonspecific DNA binding until β clamp interaction neutralizes the negative charge. We found that there are some bacterial phyla whose MutL endonucleases lack the β motif. For example, the region corresponding to the β motif is completely missing in Aquifex aeolicus MutL, and critical amino acid residues in the β motif are not conserved in Thermus thermophilus MutL. We then revealed the 1.35 Å-resolution crystal structure of A. aeolicus MutL CTD, which lacks the β motif but retains the metal-binding site for the endonuclease activity. Importantly, there was no negatively charged cluster on its surface. It was confirmed that CTDs of β motif-lacking MutLs, A. aeolicus MutL and T. thermophilus MutL, efficiently incise DNA even in the absence of β-clamp and that β-clamp shows no detectable enhancing effect on their activity. In contrast, CTD of Streptococcus mutans, a β motif-containing MutL, required β-clamp for the digestion of DNA. We propose that MutL endonucleases are divided into three subfamilies on the basis of their structural features and dependence on β-clamp.

Indispensable residue for uridine binding in the uridine-cytidine kinase family

Uridine-cytidine kinase (UCK), including human UCK2, are a family of enzymes that generally phosphorylate both uridine and cytidine. However, UCK of Thermus thermophilus HB8 (ttCK) phosphorylates only cytidine. This cytidine-restricted activity is thought to depend on Tyr93, although the precise mechanism remains unresolved. Exhaustive mutagenesis of Tyr93 in ttCK revealed that the uridine phosphorylation activity was restored only by replacement of Tyr93 with His or Gln. Replacement of His117 in human UCK2, corresponding to residue Tyr93 in ttCK, by Tyr resulted in a loss of uridine phosphorylation activity. These findings indicated that uridine phosphorylation activity commonly depends on a single residue in the UCK family.

Small-angle X-ray scattering analysis reveals the ATP-bound monomeric state of the ATPase domain from the homodimeric MutL endonuclease, a GHKL phosphotransferase superfamily protein

DNA mismatch repair is an excision system that removes mismatched bases chiefly generated by replication errors. In this system, MutL endonucleases direct the excision reaction to the error-containing strand of the duplex by specifically incising the newly synthesized strand. Both bacterial homodimeric and eukaryotic heterodimeric MutL proteins belong to the GHKL ATPase/kinase superfamily that comprises the N-terminal ATPase and C-terminal dimerization regions. Generally, the GHKL proteins show large ATPase cycle-dependent conformational changes, including dimerization-coupled ATP binding of the N-terminal domain. Interestingly, the ATPase domain of human PMS2, a subunit of the MutL heterodimer, binds ATP without dimerization. The monomeric ATP-bound state of the domain has been thought to be characteristic of heterodimeric GHKL proteins. In this study, we characterized the ATP-bound state of the ATPase domain from the Aquifex aeolicus MutL endonuclease, which is a homodimeric GHKL protein unlike the eukaryotic MutL. Gel filtration, dynamic light scattering, and small-angle X-ray scattering analyses clearly showed that the domain binds ATP in a monomeric form despite its homodimeric nature. This indicates that the uncoupling of dimerization and ATP binding is a common feature among bacterial and eukaryotic MutL endonucleases, which we suggest is closely related to the molecular mechanisms underlying mismatch repair.

The Lon protease-like domain in the bacterial RecA paralog RadA is required for DNA binding and repair

Homologous recombination (HR) plays an essential role in the maintenance of genome integrity. RecA/Rad51 paralogs have been recognized as an important factor of HR. Among them, only one bacterial RecA/Rad51 paralog, RadA, is involved in HR as an accessory factor of RecA recombinase. RadA has a unique Lon protease-like domain (LonC) at its C terminus, in addition to a RecA-like ATPase domain. Unlike Lon protease, RadAs LonC domain does not show protease activity but is still essential for RadA-mediated DNA repair. Reconciling these two facts has been difficult because RadAs tertiary structure and molecular function are unknown. Here, we describe the hexameric ring structure of RadAs LonC domain, as determined by X-ray crystallography. The structure revealed the two positively charged regions unique to the LonC domain of RadA are located at the intersubunit cleft and the central hole of a hexameric ring. Surprisingly, a functional domain analysis demonstrated the LonC domain of RadA binds DNA, with site-directed mutagenesis showing that the two positively charged regions are critical for this DNA-binding activity. Interestingly, only the intersubunit cleft was required for the DNA-dependent stimulation of ATPase activity of RadA, and at least the central hole was essential for DNA repair function. Our data provide the structural and functional features of the LonC domain and their function in RadA-mediated DNA repair.

Archaeal MutS5 tightly binds to Holliday junction similarly to eukaryotic MutSγ

Archaeal DNA recombination mechanism and the related proteins are similar to those in eukaryotes. However, no functional homolog of eukaryotic MutSγ, which recognizes Holliday junction to promote homologous recombination, has been identified in archaea. Hence, the whole molecular mechanism of archaeal homologous recombination has not yet been revealed. In this study, to identify the archaeal functional homolog of MutSγ, we focused on a functionally uncharacterized MutS homolog, MutS5, from a hyperthermophilic archaeon Pyrococcus horikoshii (phMutS5). Archaeal MutS5 has a Walker ATPase motif‐containing amino acid sequence that shows similarity to the ATPase domain of MutSγ. It is known that the ATPase domain of MutS homologs is also a dimerization domain. Chemical cross‐linking revealed that purified phMutS5 has an ability to dimerize in solution. phMutS5 bound to Holliday junction with a higher affinity than to other branched and linear DNAs, which resembles the DNA‐binding specificities of MutSγ and bacterial MutS2, a Holliday junction‐resolving MutS homolog. However, phMutS5 has no nuclease activity against branched DNA unlike MutS2. The ATPase activity of phMutS5 was significantly stimulated by the presence of Holliday junction similarly to MutSγ. Furthermore, site‐directed mutagenesis revealed that the ATPase activity is dependent on the Walker ATPase motif of the protein. These results suggest that archaeal MutS5 should stabilize the Holliday junction and play a role in homologous recombination, which is analogous to the function of eukaryotic MutSγ.

Crystal structure and DNA-binding property of the ATPase domain of bacterial mismatch repair endonuclease MutL from Aquifex aeolicus

DNA mismatch repair (MMR) system corrects mismatched bases that are generated mainly by DNA replication errors. The repair system excises the error-containing single-stranded region and enables the re-synthesis of the strand. In the early reactions of MMR, MutL endonuclease incises the newly-synthesized/error-containing strand of the duplex to initiate the downstream excision reaction. MutL endonuclease consists of the N-terminal ATPase and C-terminal endonuclease domains. In this study, we report the crystal structure of the ATPase domain of MutL endonuclease from Aquifex aeolicus. The overall structure of the domain was similar to those of human MutL homologs and Escherichia coli MutL, although E. coli MutL has no endonuclease activity. The ATPase domain was comprised of two subdomains: the N-terminal ATP-binding subdomain and the C-terminal α-β sandwich subdomain. Site-directed mutagenesis experiment identified DNA-interacting eight basic amino acid residues, which were distributed across both the two subdomains and formed a DNA-binding cleft. Docking simulation between the structures of the ATPase and endonuclease domains generated a reliable model structure for the full-length A. aeolicus MutL, which satisfies our previous result of small-angle X-ray scattering analysis. On the basis of the model structure and further experimental results, we concluded that the two separate DNA-binding sites in the full-length A. aeolicus MutL simultaneously bind a dsDNA molecule.

Structural Insights into l-Tryptophan Dehydrogenase from a Photoautotrophic Cyanobacterium, Nostoc punctiforme

ABSTRACT l-Tryptophan dehydrogenase from Nostoc punctiforme NIES-2108 (NpTrpDH), despite exhibiting high amino acid sequence identity (>30%)/homology (>50%) with NAD(P)+-dependent l-Glu/l-Leu/l-Phe/l-Val dehydrogenases, exclusively catalyzes reversible oxidative deamination of l-Trp to 3-indolepyruvate in the presence of NAD+. Here, we determined the crystal structure of the apo form of NpTrpDH. The structure of the NpTrpDH monomer, which exhibited high similarity to that of l-Glu/l-Leu/l-Phe dehydrogenases, consisted of a substrate-binding domain (domain I, residues 3 to 133 and 328 to 343) and an NAD+/NADH-binding domain (domain II, residues 142 to 327) separated by a deep cleft. The apo-NpTrpDH existed in an open conformation, where domains I and II were apart from each other. The subunits dimerized themselves mainly through interactions between amino acid residues around the β-1 strand of each subunit, as was observed in the case of l-Phe dehydrogenase. The binding site for the substrate l-Trp was predicted by a molecular docking simulation and validated by site-directed mutagenesis. Several hydrophobic residues, which were located in the active site of NpTrpDH and possibly interacted with the side chain of the substrate l-Trp, were arranged similarly to that found in l-Leu/l-Phe dehydrogenases but fairly different from that of an l-Glu dehydrogenase. Our crystal structure revealed that Met-40, Ala-69, Ile-74, Ile-110, Leu-288, Ile-289, and Tyr-292 formed a hydrophobic cluster around the active site. The results of the site-directed mutagenesis experiments suggested that the hydrophobic cluster plays critical roles in protein folding, l-Trp recognition, and catalysis. Our results provide critical information for further characterization and engineering of this enzyme. IMPORTANCE In this study, we determined the three-dimensional structure of l-Trp dehydrogenase, analyzed its various site-directed substitution mutants at residues located in the active site, and obtained the following informative results. Several residues in the active site form a hydrophobic cluster, which may be a part of the hydrophobic core essential for protein folding. To our knowledge, there is no previous report demonstrating that a hydrophobic cluster in the active site of any l-amino acid dehydrogenase may have a critical impact on protein folding. Furthermore, our results suggest that this hydrophobic cluster could strictly accommodate l-Trp. These studies show the structural characteristics of l-Trp dehydrogenase and hence would facilitate novel applications of l-Trp dehydrogenase.